A Novel Improved Particle Swarm Optimization With Long-Short Term Memory Hybrid Model for Stock Indices Forecast

Stock market volatility has a significant impact on many economic and financial activities in the world. Forecasting stock price movement plays an important role setting investment strategy or determining right timing for trading. However, movements are noisy, nonlinear, chaotic. It is difficult to forecast trends improving return investment. Here, we proposed novel improved particle swarm optimization (IPSO) long-short term memory (LSTM) hybrid model forecasting. An adaptive mutation factor was used as parameter avoid premature convergence local optimum. Furthermore, presented nonlinear approach improve inertia weight of then IPSO optimize hyperparameters LSTM. The experimental results showed that outperformed other related baseline models: support-vector regression, LSTM PSO-LSTM Australian index. These indicated possesses high reliability good forecasting capability.